Merge pull request #34 from common-workflow-language/unscoped_ids

Unscoped ids

Author: mr-c

schema_salad/jsonld_context.py

@@ -175,3 +175,52 @@ def salad_to_jsonld_context(j, schema_ctx):
         process_type(t, g, context, defaultBase, namespaces, defaultPrefix)
 
     return (context, g)
+
+def fix_jsonld_ids(obj, ids):
+    # type: (Union[Dict[unicode, Any], List[Dict[unicode, Any]]], List[unicode]) -> None
+    if isinstance(obj, dict):
+        for i in ids:
+            if i in obj:
+                obj["@id"] = obj[i]
+        for v in obj.values():
+            fix_jsonld_ids(v, ids)
+    if isinstance(obj, list):
+        for entry in obj:
+            fix_jsonld_ids(entry, ids)
+
+def makerdf(workflow, wf, ctx):
+    # type: (Union[str, unicode], Union[List[Dict[unicode, Any]], Dict[unicode, Any]], Loader.ContextType) -> Graph
+    prefixes = {}
+    idfields = []
+    for k,v in ctx.iteritems():
+        if isinstance(v, dict):
+            url = v["@id"]
+        else:
+            url = v
+        if url == "@id":
+            idfields.append(k)
+        doc_url, frg = urlparse.urldefrag(url)
+        if "/" in frg:
+            p, _ = frg.split("/")
+            prefixes[p] = u"%s#%s/" % (doc_url, p)
+
+    if isinstance(wf, list):
+        wf = {
+            "@context": ctx,
+            "@graph": wf
+        }
+    else:
+        wf["@context"] = ctx
+
+    fix_jsonld_ids(wf, idfields)
+
+    g = Graph().parse(data=json.dumps(wf), format='json-ld', location=workflow)
+
+    # Bug in json-ld loader causes @id fields to be added to the graph
+    for s,p,o in g.triples((None, URIRef("@id"), None)):
+        g.remove((s, p, o))
+
+    for k2,v2 in prefixes.iteritems():
+        g.namespace_manager.bind(k2, v2)
+
+    return g

schema_salad/main.py

@@ -23,9 +23,8 @@
 
 
 def printrdf(workflow, wf, ctx, sr):
-    # type: (str, Union[List, Dict[Any, Any], str, unicode], Dict[unicode, Any], str) -> None
-    g = Graph().parse(data=json.dumps(wf), format='json-ld',
-                      location=workflow, context=ctx)
+    # type: (str, Union[List[Dict[unicode, Any]], Dict[unicode, Any]], Dict[unicode, Any], str) -> None
+    g = jsonld_context.makerdf(workflow, wf, ctx)
     print(g.serialize(format=sr))
 
 
@@ -210,8 +209,12 @@ def main(argsl=None):  # type: (List[str]) -> int
 
     # Optionally convert the document to RDF
     if args.print_rdf:
-        printrdf(args.document, document, schema_ctx, args.rdf_serializer)
-        return 0
+        if isinstance(document, (dict, list)):
+            printrdf(args.document, document, schema_ctx, args.rdf_serializer)
+            return 0
+        else:
+            print("Document must be a dictionary or list.")
+            return 1
 
     if args.print_metadata:
         print(json.dumps(doc_metadata, indent=4))

schema_salad/metaschema/metaschema_base.yml

@@ -43,19 +43,19 @@ $graph:
   type: record
   doc: A field of a record.
   fields:
-    - name: name
+    name:
       type: string
       jsonldPredicate: "@id"
       doc: |
         The name of the field
 
-    - name: doc
+    doc:
       type: string?
       doc: |
         A documentation string for this field
       jsonldPredicate: "sld:doc"
 
-    - name: type
+    type:
       type:
         - PrimitiveType
         - RecordSchema
@@ -81,7 +81,7 @@ $graph:
 - name: RecordSchema
   type: record
   fields:
-    - name: type
+    type:
       doc: "Must be `record`"
       type:
         name: Record_symbol
@@ -93,9 +93,12 @@ $graph:
         _type: "@vocab"
         typeDSL: true
         refScope: 2
-    - name: "fields"
+    fields:
       type: RecordField[]?
-      jsonldPredicate: "sld:fields"
+      jsonldPredicate:
+        _id: sld:fields
+        mapSubject: name
+        mapPredicate: type
       doc: "Defines the fields of the record."
 
 
@@ -104,7 +107,7 @@ $graph:
   doc: |
     Define an enumerated type.
   fields:
-    - name: type
+    type:
       doc: "Must be `enum`"
       type:
         name: Enum_symbol
@@ -116,7 +119,7 @@ $graph:
         _type: "@vocab"
         typeDSL: true
         refScope: 2
-    - name: "symbols"
+    symbols:
       type: string[]
       jsonldPredicate:
         _id: "sld:symbols"
@@ -128,7 +131,7 @@ $graph:
 - name: ArraySchema
   type: record
   fields:
-    - name: type
+    type:
       doc: "Must be `array`"
       type:
         name: Array_symbol
@@ -140,7 +143,7 @@ $graph:
         _type: "@vocab"
         typeDSL: true
         refScope: 2
-    - name: items
+    items:
       type:
         - PrimitiveType
         - RecordSchema

schema_salad/ref_resolver.py

@@ -135,8 +135,9 @@ def expand_url(self, url, base_url, scoped_id=False, vocab_term=False, scoped_re
                 frg = splitbase.fragment + u"/" + split.path
             else:
                 frg = split.path
+            pt = splitbase.path if splitbase.path else "/"
             url = urlparse.urlunsplit(
-                (splitbase.scheme, splitbase.netloc, splitbase.path, splitbase.query, frg))
+                (splitbase.scheme, splitbase.netloc, pt, splitbase.query, frg))
         elif scoped_ref is not None and not split.fragment:
             pass
         else:

schema_salad/schema.py

@@ -97,7 +97,11 @@ def get_metaschema():
             "@type": "@id",
             "refScope": 1
         },
-        "fields": "sld:fields",
+        "fields": {
+            "@id": "sld:fields",
+            "mapSubject": "name",
+            "mapPredicate": "type"
+        },
         "float": "http://www.w3.org/2001/XMLSchema#float",
         "identity": "https://w3id.org/cwl/salad#JsonldPredicate/identity",
         "int": "http://www.w3.org/2001/XMLSchema#int",

setup.py

@@ -41,7 +41,7 @@
 extras_require = {}               # TODO: to be removed when the above is added
 
 setup(name='schema-salad',
-      version='1.12',
+      version='1.13',
       description='Schema Annotations for Linked Avro Data (SALAD)',
       long_description=open(README).read(),
       author='Common workflow language working group',

tests/test_examples.py

@@ -2,8 +2,11 @@
 import schema_salad.ref_resolver
 import schema_salad.main
 import schema_salad.schema
+from schema_salad.jsonld_context import makerdf
 import rdflib
 import ruamel.yaml as yaml
+import json
+
 try:
     from ruamel.yaml import CSafeLoader as SafeLoader
 except ImportError:
@@ -249,5 +252,75 @@ def test_typedsl_ref(self):
         ra, _ = ldr.resolve_all({"type": "File[]?"}, "")
         self.assertEqual({'type': ['null', {'items': 'File', 'type': 'array'}]}, ra)
 
+    def test_scoped_id(self):
+        ldr = schema_salad.ref_resolver.Loader({})
+        ctx = {
+            "id": "@id",
+            "location": {
+                "@id": "@id",
+                "@type": "@id"
+            },
+            "bar": "http://example.com/bar",
+            "ex": "http://example.com/"
+        }
+        ldr.add_context(ctx)
+
+        ra, _ = ldr.resolve_all({
+            "id": "foo",
+            "bar": {
+                "id": "baz"
+            }
+        }, "http://example.com")
+        self.assertEqual({'id': 'http://example.com/#foo',
+                          'bar': {
+                              'id': 'http://example.com/#foo/baz'},
+                      }, ra)
+
+        g = makerdf(None, ra, ctx)
+        print(g.serialize(format="n3"))
+
+        ra, _ = ldr.resolve_all({
+            "location": "foo",
+            "bar": {
+                "location": "baz"
+            }
+        }, "http://example.com", checklinks=False)
+        self.assertEqual({'location': 'http://example.com/foo',
+                          'bar': {
+                              'location': 'http://example.com/baz'},
+                      }, ra)
+
+        g = makerdf(None, ra, ctx)
+        print(g.serialize(format="n3"))
+
+        ra, _ = ldr.resolve_all({
+            "id": "foo",
+            "bar": {
+                "location": "baz"
+            }
+        }, "http://example.com", checklinks=False)
+        self.assertEqual({'id': 'http://example.com/#foo',
+                          'bar': {
+                              'location': 'http://example.com/baz'},
+                      }, ra)
+
+        g = makerdf(None, ra, ctx)
+        print(g.serialize(format="n3"))
+
+        ra, _ = ldr.resolve_all({
+            "location": "foo",
+            "bar": {
+                "id": "baz"
+            }
+        }, "http://example.com", checklinks=False)
+        self.assertEqual({'location': 'http://example.com/foo',
+                          'bar': {
+                              'id': 'http://example.com/#baz'},
+                      }, ra)
+
+        g = makerdf(None, ra, ctx)
+        print(g.serialize(format="n3"))
+
+
 if __name__ == '__main__':
     unittest.main()